Science Time Series: Deep Learning in Hydrology

This research is part of a systematic study of scientific time series. In the last three years, hundreds of papers and over fifty new deep-learning models have been described for time series models. These mainly focus on the key aspect of time dependence, whereas in some scientific time series, the situation is more complex with multiple locations, each location having multiple observed and target time-dependent streams and multiple exogenous (known) properties that are either constant or time-dependent. Here, we analyze the hydrology time series using the CAMELS and Caravan global datasets on catchment rainfall and runoff. Together, these have up to 6 observed streams and up to 209 static parameters defined at each of about 8000 locations. This analysis is fully open source with a Jupyter Notebook running on Google Colab for both an LSTM-based analysis and the data engineering preprocessing. Our goal is to investigate the importance of exogenous data, which we look at using eight different choices on representative hydrology tasks. Increasing the exogenous information significantly improves the data representation, with the mean square error decreasing to 60% of its initial value in the largest dataset examined. We present the initial results of studies of other deep-learning neural network architectures where the approaches that can use the full observed and exogenous observations outperform less flexible methods, including Foundation models. Using the natural annual periodic exogenous time series produces the largest impact, but the static and other periodic exogenous streams are also important. Our analysis is intended to be valuable as an educational resource and benchmark.

Study of Dropout in PointPillars with 3D Object Detection

3D object detection is critical for autonomous driving, leveraging deep learning techniques to interpret LiDAR data. The PointPillars architecture is a prominent model in this field, distinguished by its efficient use of LiDAR data. This study provides an analysis of enhancing the performance of PointPillars model under various dropout rates to address overfitting and improve model generalization. Dropout, a regularization technique, involves randomly omitting neurons during training, compelling the network to learn robust and diverse features. We systematically compare the effects of different enhancement techniques on the model's regression performance during training and its accuracy, measured by Average Precision (AP) and Average Orientation Similarity (AOS). Our findings offer insights into the optimal enhancements, contributing to improved 3D object detection in autonomous driving applications.

Time Series Foundation Models and Deep Learning Architectures for Earthquake Temporal and Spatial Nowcasting

Advancing the capabilities of earthquake nowcasting, the real-time forecasting of seismic activities remains a crucial and enduring objective aimed at reducing casualties. This multifaceted challenge has recently gained attention within the deep learning domain, facilitated by the availability of extensive, long-term earthquake datasets. Despite significant advancements, existing literature on earthquake nowcasting lacks comprehensive evaluations of pre-trained foundation models and modern deep learning architectures. These architectures, such as transformers or graph neural networks, uniquely focus on different aspects of data, including spatial relationships, temporal patterns, and multi-scale dependencies. This paper addresses the mentioned gap by analyzing different architectures and introducing two innovation approaches called MultiFoundationQuake and GNNCoder. We formulate earthquake nowcasting as a time series forecasting problem for the next 14 days within 0.1-degree spatial bins in Southern California, spanning from 1986 to 2024. Earthquake time series is forecasted as a function of logarithm energy released by quakes. Our comprehensive evaluation employs several key performance metrics, notably Nash-Sutcliffe Efficiency and Mean Squared Error, over time in each spatial region. The results demonstrate that our introduced models outperform other custom architectures by effectively capturing temporal-spatial relationships inherent in seismic data. The performance of existing foundation models varies significantly based on the pre-training datasets, emphasizing the need for careful dataset selection. However, we introduce a new general approach termed MultiFoundationPattern that combines a bespoke pattern with foundation model results handled as auxiliary streams. In the earthquake case, the resultant MultiFoundationQuake model achieves the best overall performance.

Feasibility Study on Active Learning of Smart Surrogates for Scientific Simulations

High-performance scientific simulations, important for comprehension of complex systems, encounter computational challenges especially when exploring extensive parameter spaces. There has been an increasing interest in developing deep neural networks (DNNs) as surrogate models capable of accelerating the simulations. However, existing approaches for training these DNN surrogates rely on extensive simulation data which are heuristically selected and generated with expensive computation -- a challenge under-explored in the literature. In this paper, we investigate the potential of incorporating active learning into DNN surrogate training. This allows intelligent and objective selection of training simulations, reducing the need to generate extensive simulation data as well as the dependency of the performance of DNN surrogates on pre-defined training simulations. In the problem context of constructing DNN surrogates for diffusion equations with sources, we examine the efficacy of diversity- and uncertainty-based strategies for selecting training simulations, considering two different DNN architecture. The results set the groundwork for developing the high-performance computing infrastructure for Smart Surrogates that supports on-the-fly generation of simulation data steered by active learning strategies to potentially improve the efficiency of scientific simulations.

A Comprehensive Evaluation of Generative Models in Calorimeter Shower Simulation

The pursuit of understanding fundamental particle interactions has reached unparalleled precision levels. Particle physics detectors play a crucial role in generating low-level object signatures that encode collision physics. However, simulating these particle collisions is a demanding task in terms of memory and computation which will be exasperated with larger data volumes, more complex detectors, and a higher pileup environment in the High-Luminosity LHC. The introduction of "Fast Simulation" has been pivotal in overcoming computational bottlenecks. The use of deep-generative models has sparked a surge of interest in surrogate modeling for detector simulations, generating particle showers that closely resemble the observed data. Nonetheless, there is a pressing need for a comprehensive evaluation of their performance using a standardized set of metrics. In this study, we conducted a rigorous evaluation of three generative models using standard datasets and a diverse set of metrics derived from physics, computer vision, and statistics. Furthermore, we explored the impact of using full versus mixed precision modes during inference. Our evaluation revealed that the CaloDiffusion and CaloScore generative models demonstrate the most accurate simulation of particle showers, yet there remains substantial room for improvement. Our findings identified areas where the evaluated models fell short in accurately replicating Geant4 data.

RINAS: Training with Dataset Shuffling Can Be General and Fast

Deep learning datasets are expanding at an unprecedented pace, creating new challenges for data processing in model training pipelines. A crucial aspect of these pipelines is dataset shuffling, which significantly improves unbiased learning and convergence accuracy by adhering to the principles of random sampling. However, loading shuffled data for large datasets incurs significant overhead in the deep learning pipeline and severely impacts the end-to-end training throughput. To mitigate this, current deep learning systems often resort to partial dataset shuffling, sacrificing global randomness to maintain acceptable training throughput on large datasets, still leaving global shuffling efficiency issues not fully explored. In this work, we present RINAS, a data loading framework that systematically addresses the performance bottleneck of loading global shuffled datasets. Our key contribution is to offer an intra-batch unordered data fetching approach, which unleashes unexplored parallelism of data loading. We implement RINAS under the PyTorch framework for common dataset libraries HuggingFace and TorchVision. Our experimental results show that RINAS improves the throughput of general language model training and vision model training by up to 59% and 89%, respectively.

RTP: Rethinking Tensor Parallelism with Memory Deduplication

In the evolving landscape of neural network models, one prominent challenge stand out: the significant memory overheads associated with training expansive models. Addressing this challenge, this study delves deep into the Rotated Tensor Parallelism (RTP). RTP is an innovative approach that strategically focuses on memory deduplication in distributed training environments. It boasts of unique features like a customized communication primitive and the Flyweight Pattern initialization. Furthermore, RTP ensures a seamless overlap between partition computation and partition weight communication, optimizing the training process. Our empirical evaluations underscore RTP's efficiency, revealing that its memory consumption during distributed system training is remarkably close to the optimal - distributing the memory overhead of a single machine equitably among multiple machines. The experimental results demonstrate that RTP is capable of achieving comparable performance to Distributed Data Parallel while providing support for significantly larger models with near-linear scalability in terms of memory. Code of RTP is available at https://github.com/wdlctc/rtp.

In-depth Analysis On Parallel Processing Patterns for High-Performance Dataframes

The Data Science domain has expanded monumentally in both research and industry communities during the past decade, predominantly owing to the Big Data revolution. Artificial Intelligence (AI) and Machine Learning (ML) are bringing more complexities to data engineering applications, which are now integrated into data processing pipelines to process terabytes of data. Typically, a significant amount of time is spent on data preprocessing in these pipelines, and hence improving its e fficiency directly impacts the overall pipeline performance. The community has recently embraced the concept of Dataframes as the de-facto data structure for data representation and manipulation. However, the most widely used serial Dataframes today (R, pandas) experience performance limitations while working on even moderately large data sets. We believe that there is plenty of room for improvement by taking a look at this problem from a high-performance computing point of view. In a prior publication, we presented a set of parallel processing patterns for distributed dataframe operators and the reference runtime implementation, Cylon [1]. In this paper, we are expanding on the initial concept by introducing a cost model for evaluating the said patterns. Furthermore, we evaluate the performance of Cylon on the ORNL Summit supercomputer.

Analyzing the Performance of Deep Encoder-Decoder Networks as Surrogates for a Diffusion Equation

Neural networks (NNs) have proven to be a viable alternative to traditional direct numerical algorithms, with the potential to accelerate computational time by several orders of magnitude. In the present paper we study the use of encoder-decoder convolutional neural network (CNN) as surrogates for steady-state diffusion solvers. The construction of such surrogates requires the selection of an appropriate task, network architecture, training set structure and size, loss function, and training algorithm hyperparameters. It is well known that each of these factors can have a significant impact on the performance of the resultant model. Our approach employs an encoder-decoder CNN architecture, which we posit is particularly well-suited for this task due to its ability to effectively transform data, as opposed to merely compressing it. We systematically evaluate a range of loss functions, hyperparameters, and training set sizes. Our results indicate that increasing the size of the training set has a substantial effect on reducing performance fluctuations and overall error. Additionally, we observe that the performance of the model exhibits a logarithmic dependence on the training set size. Furthermore, we investigate the effect on model performance by using different subsets of data with varying features. Our results highlight the importance of sampling the configurational space in an optimal manner, as this can have a significant impact on the performance of the model and the required training time. In conclusion, our results suggest that training a model with a pre-determined error performance bound is not a viable approach, as it does not guarantee that edge cases with errors larger than the bound do not exist. Furthermore, as most surrogate tasks involve a high dimensional landscape, an ever increasing training set size is, in principle, needed, however it is not a practical solution.

GTrans: Spatiotemporal Autoregressive Transformer with Graph Embeddings for Nowcasting Extreme Events

Spatiotemporal time series nowcasting should preserve temporal and spatial dynamics in the sense that generated new sequences from models respect the covariance relationship from history. Conventional feature extractors are built with deep convolutional neural networks (CNN). However, CNN models have limits to image-like applications where data can be formed with high-dimensional arrays. In contrast, applications in social networks, road traffic, physics, and chemical property prediction where data features can be organized with nodes and edges of graphs. Transformer architecture is an emerging method for predictive models, bringing high accuracy and efficiency due to attention mechanism design. This paper proposes a spatiotemporal model, namely GTrans, that transforms data features into graph embeddings and predicts temporal dynamics with a transformer model. According to our experiments, we demonstrate that GTrans can model spatial and temporal dynamics and nowcasts extreme events for datasets. Furthermore, in all the experiments, GTrans can achieve the highest F1 and F2 scores in binary-class prediction tests than the baseline models.